UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 8

1

2

Kigali, Rwanda

3

E-mail:

aristide.maniriho@uliege.be,

musabanganji@gmail.com,

fnkikabahizi@gmail.com,

cruranga@gmail.com,

poverty in the developing world (Deaton, 1997). Households act in housing and allocate their disposable income to

current consumption or for savings and future consumer needs (Cismas et al., 2010). Using consumption growth to

measure material wealth follows the argument that income likely underestimates the material wealth of households in

the context of developing countries (Deaton & Zaidi, 2002). At the household level, the total expenditure expresses the

use of revenues, which are normally used for consumption or for household savings. Total expenditures include money

spending regardless of its destination (consumption, taxes and binding payments, birds and animals acquisitions,

buildings and land, other investment costs, etc.), the value of benefits (goods and services) free or reduced price

evaluated at the sale price of the bidding part, as well as the value of food and non-food consumption from own

resources determined from the monthly average prices of those products (Cismas et al., 2010). Consumption has been

chosen as a good indicator of household wealth because the current income is typically vulnerable to temporary

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 9

fluctuations due to factors such as layoffs or changes in family status which cause current income to vary more than

consumption (Cutler & Katz, 1991). The main components of household consumption expenditures are food or non-

food items, services, and transfers to public and private administration and to social security budgets in the form of

taxes, levies, contributions, and coverage of domestic production related needs. Another component is an expenditure

incurred for the portion of food and drink purchased for consumption, which are not consumed in the reference period,

(McConnell & Bruce, 2005; Dwivedi, 2006; Schiller, 2006; Nicholson & Snyder, 2011; Perloff, 2008; Varian, 2010;

Besanko & Braeutigam, 2011; Griffiths & Wall, 2011; Samuelson & Marks, 2012). Specifically for households, their

demand is influenced by several factors such as income, education, age, gender (of the household head and other

household members), weather conditions, location, tastes and preferences as well as the household size (among other

factors).

The size of food expenditures can be used as a welfare indicator (Zimmerman, 1932). It has been observed that in

poor households, the share of household income allocated to food stuffs increases with the increase in income (Maki &

Ohira, 2014; Maki & Kamwe, 2012). A household is food secure when the Engel coefficient for food reaches its peak

and starts declining (Maki & Ohira, 2014). It was also observed that when the standard of living is still low, the

household’s need is to satisfy the most urgent needs of living, namely food, clothing and home, and in this line,

Heshmati and Rashidghalam (2018) identified the differences between consumption and income measures of poverty

in Rwanda. By means of income, head count absolute poverty is above 0.80 in all districts, whereas the absolute poverty

by means of consumption is 0.531 on average. The question comes hereby to know the sources of different trends in

income and consumption and their implications for food security across districts and regions in Rwanda.

It is expected that the results of this study will be used by policy makers, the heads of households while planning for

household expenditures, and social workers while sensitizing communities on welfare improvements. Knowing the main

determinants of household expenditures, decision makers know where more efforts are needed to improve household

welfare conditions. Similarly, households are informed about the factors that affect their consumption decisions and

how they learn how to deal with each factor to achieve high standards of living. In regards to researchers, this study

contributes to the set of knowledge related to household welfare economics in Rwanda.

Figure 1. Comparison between GDP per capita, income per capita, and expenditure per capita in Rwanda (1996-2016).

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 10

This study used the Comprehensive Food Security and Vulnerability Analysis (CFSVA) survey data collected

from a random sample of 7,500 households at national level in 2015 in Rwanda.

estimate the price of basic food items at the district level. Fresh beans, fresh maize, banana for cooking, banana

fruits, green vegetables, cassava for cooking, cassava for flour, cassava leaves, and avocado are among the nine

food products herewith considered. Since a large number of variables were being dealt with, a principal

component indices to minimize data dimensions and cope with multicollinearity without losing any variables

was computed: the socioeconomic index was calculated using household socioeconomic characteristics (age,

heteroskedasticity that is common in most cross-sectional data (Wooldridge, 2013). The linear regression

model for examining household spending was developed according to Gujarati and Porter (2010) as per the

Equation (1).

∑

β

k

X

+ ε

i

(1)

where C stands for the monthly household expenditures while the subscript I indexes a sampled household; j

is the number of observations of an individual household I for each dependent variable; Xs denote the vectors

of independent variables; β represents the vector coefficients; the intercept α represents the expected value of

For the purposes of this study, three household demand functions have been estimated: total consumption

in developing countries (Dercon et al., 2009; Islam & Maitra, 2012). All variables selected for this study are

described in Table 1.

Table 1. Definitions and descriptive statistics of study variables.

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 11

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 12

positive impact on household expenditures. This means that as these parameters, indices, and variables rise, household

expenditures rise as well. Poverty and the distance index, on the other hand, have a negative but significant impact on

household spending. The results reveal that while price has a positive and substantial impact on household food

expenditures, it has a negative but substantial impact on non-food expenditures. Surprisingly, remittances have a negative

and large impact on household expenses.

head, poverty status, and distance to the nearest essential amenities (all-weather road, health facility) were

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 13

community facilities (road, health facility) were reported among the primary determinants with negative effect

Table 3. Robustness test. The OLS estimates.

The dependent variable is the total household expenditures for poor and non-poor households, respectively.

Note: *, **, and *** mean that the estimated coefficient is significant at the level of 10 per cent, 5 per cent, and 1 per cent, respectively. The

dependent variable is log transformed.

a

The reported R-square is considered significant as cross-section data are used (Wooldridge, 2002).

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 14

index (age, gender, education, household size, marital situation, and polygamy status), the number of livestock units, the

share of non-farm income, and the livelihood zone where the household operates all have a positive impact on the

likelihood of a household consuming protein. The household's poverty status, as well as the distance to basic

infrastructure, have a negative impact on this probability.

Table 4. Additional robustness test.

*, **, and *** mean that the estimated coefficient is significant at the level of 10 per cent, 5 per cent, and 1 per cent, respectively.

3.3. Discussion of the findings

The main results from econometric estimations in Tables 2 and A1 show that the socioeconomic characteristics are very

important drivers of household expenditures (here the household size, the age of the household head, and education

level of the household head). This finding is in line with the findings of Donkoh and Amikuzuno (2011), Umeh and

Asogwa (2012) and Nilsson et al. (2019) who reported the age and education of the household head as well as the

household size

among the determinants of household expenditures but is in contrast with Davis et al.’s (1983) finding

that the education has no significant impact on household education expenditure. These results are reflective of the vital

importance of household control factors in their demand decisions.

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 15

The findings also revealed that having productive assets (livestock units, land size) and financial resources (access to

credit, non-farm income share) have a substantial impact on household demand. This result supports the findings that

the land size (Umeh & Asogwa, 2012), possession of durable assets (Donkoh & Amikuzuno, 2011), the house size

(Hussain and Asad, 2012), the household income and wealth (Davis et al., 1983; Khan & Abdullah, 2010; Umeh &

Asogwa, 2012; Wang et al., 2016) and the number of livestock units (Nilsson et al., 2019) are among the factors that

The results show that as the average price of food products rise, household food expenditures rise as well, a finding

that is consistent with Cismas et al.’s (2010) assertion that the average price is one of the primary factors determining

the value of food and non-food products consumed in households. The negative effect of average food prices on non-

food expenditures is explained by cross price elasticity (Schotter, 2008; Varian, 2010; Besanko & Braeutigam, 2011),

which states that an increase in food products (or necessities) causes households to reduce non-food product

consumption to maintain the same level of food consumption.

Another critical finding is the effect of locational factors specifically the distance to a health facility, distance to a

market and the household location in an urban area on the household consumption decisions. This follows the finding

on the diversification of livelihood sources, and Nilsson et al. (2019) who underlined the negative and significant effect

of the distance to asphalt road on household consumption expenditures.

Consumption expenditure has been a good indicator of household wealth and welfare conditions especially in

developing countries. The salient argument behind this is that income measures tend to underestimate the material

wealth of households. The analysis of household consumption expenditures has interested a significant number of

researchers all around the world, but it has not been the case in Rwanda where the documentation on the factors

impacting household consumption decisions is still limited. This study attempted to identify the determinants of

household consumption expenditures in Rwanda. A multivariate regressions model was specified, and three demand

functions were hereby estimated -- total household demand function, food demand function and non-food demand

household expenditures, while the distance to improved road, a market and a health facility negatively affects the

household expenditures, which means that household consumption is higher in urban rather than in rural areas. In

addition to this, it was reported that the number of livelihood activities done by household members, land consolidation

and maize production affect positively household demand decisions, while the distance to all weather roads and a health

facility have negative effects on household consumption expenditures.

The findings show that the most significant factors affecting household demand decisions are household size, the age

of the household head, the education level of the head, the number of livestock units, poverty situation, the access to

credits and the size of cultivated land. In addition, the results revealed that most of these factors apply to both poor and

non-poor households, as well as for households in both rural and urban areas. The logit estimates revealed that

household socioeconomic characteristics, livestock ownership, and non-farm income share are among the key variables

UKH Journal of Social Sciences | Volume 5 • Number 1 • 2021 16

References

Besanko, D. & Braeutigam, R. (2011). Microeconomics (4

th

ed). New York, USA: John Wiley and Sons, Inc. pp. 822.

Bopape, L. & Myers, R. (2007). Analysis of household demand for food in South Africa: model selection, expenditure endogeneity, and

Cismas, L., Miculescu, A., & Oţil, M. (2010). Income and Expenditure of Households in Romania. Annals of University

Cutler, D. M., Katz, L. F. (1991). Macroeconomic performance and the disadvantaged. Brookings papers on economic activity,

Davis, C. G., Moussie, M., Dinning, J. S. & Christakis, G. J. (1983). Socioeconomic determinants of food expenditure

Dercon, S., Gilligan, D. O., Hoddinott, J. & Woldehanna, T. (2009). The impact of agricultural extension and roads on

Donkoh, S. A. & Amikuzuno, J. A (2011). The determinants of household education expenditure in Ghana. Educational

Dwivedi, D. N. (2006). Microeconomics: Theory and Applications. New Delhi, India, Dorling Kindersley (India) Pvt. Ltd. pp.

Griffiths, A. & Wall, S. (2011). Economics for Business and Management (3

ed). North York, Canada: Pearson Education Ltd.

Gujarati, D. N & Porter, D. C. (2010). Essentials of Econometrics, 4

ed. New York, USA: McGraw-Hill Irwin. pp. 577.

Islam, A. & Maitra, P. (2012). Health shocks and consumption smoothing in rural households: Does microcredit have

Johnson, R. A. & Wichern, D. W. (2004). Multivariate analysis. Encyclopedia of Statistical Sciences, 8.

Johnson, R. A. & Wichern, D. W. (2015). Applied multivariate statistical analysis. Statistics, 6215(10), 10.

Keynes, J. M. (1937). The general theory of employment. The quarterly journal of economics, 51(2), 209-223.

Khan, H. H. A. & Abdullah, H. (2010). Saving determinants in Malaysia. Jurnal Ekonomi Malaysia, 44, 23-34.